From 4905e589d74e72fccb6e050d4e0db0d96c8e3930 Mon Sep 17 00:00:00 2001
From: John Hawkinson <jhawk@mit.edu>
Date: Sat, 8 Apr 2017 04:21:52 -0400
Subject: [PATCH] [kaltura] Be rigorous on iframe

Per @dstftw, don't pull out just any kaltura.com iframes, make sure they
have /p/{PARTNER_ID} and &entry_id={ENTRY_ID} and return a kaltura: URL.

Go back to specifying the IE is Kaltura in url_result().
---
 youtube_dl/extractor/generic.py |  2 +-
 youtube_dl/extractor/kaltura.py | 19 +++++++++++--------
 2 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 4bfa3f8a1..658533cf6 100644
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -2305,7 +2305,7 @@ class GenericIE(InfoExtractor):
         # Look for Kaltura embeds
         kaltura_url = KalturaIE._extract_url(webpage)
         if kaltura_url:
-            return self.url_result(smuggle_url(kaltura_url, {'source_url': url}))
+            return self.url_result(smuggle_url(kaltura_url, {'source_url': url}), KalturaIE.ie_key())
 
         # Look for Eagle.Platform embeds
         eagleplatform_url = EaglePlatformIE._extract_url(webpage)
diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py
index 741dd8dc5..6e992ee4b 100644
--- a/youtube_dl/extractor/kaltura.py
+++ b/youtube_dl/extractor/kaltura.py
@@ -128,7 +128,17 @@ class KalturaIE(InfoExtractor):
                         (?P<q2>["\'])entry_?[Ii]d(?P=q2)
                     )\s*:\s*
                     (?P<q3>["\'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3)
-                ''', webpage))
+                ''', webpage) or
+            re.search(
+                # <iframe src="http://www.kaltura.com/p/{PARTNER_ID}/sp/{PARTNER_ID}00/embedIframeJs/uiconf_id/{UICONF_ID}/partner_id/{PARTNER_ID}?iframeembed=true&playerId={UNIQUE_OBJ_ID}&entry_id={ENTRY_ID}" width="400" height="330" allowfullscreen webkitallowfullscreen mozAllowFullScreen frameborder="0"></iframe>
+                r'''(?xs)
+                    (?P<q1>["\'])
+                      (?:https?:)?//(?:www\.)?kaltura\.com/p/(?P<partner_id>\d+)/
+                      (?:(?!(?P=q1)).)*
+                      [\?&]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+)
+                    (?P=q1)
+                ''', webpage)
+        )
         if mobj:
             embed_info = mobj.groupdict()
             url = 'kaltura:%(partner_id)s:%(id)s' % embed_info
@@ -140,13 +150,6 @@ class KalturaIE(InfoExtractor):
                 url = smuggle_url(url, {'service_url': service_url.group(1)})
             return url
 
-        # Check for an iframe, which may require redirection.
-        mobj = re.search(
-            r"<iframe[^>]+src=['\"](?P<url>(https?:)?//www\.kaltura\.com/[^'\"]+)['\"]",
-            webpage)
-        if mobj:
-            return mobj.group('url')
-
     def _kaltura_api_call(self, video_id, actions, service_url=None, *args, **kwargs):
         params = actions[0]
         if len(actions) > 1: